Short-term and working memory capacity and the language device: Chunking and parsing complexity

Abstract

Many general linguistic theories and language processing frameworks have assumed that language processing is largely a chunking procedure and that it is underpinned and constrained by our memory limitations. Despite this general consensus, the distinction between short-term memory (STM) and working memory (WM) limitations as they relate to language processing has remained elusive. To resolve this issue, we propose an integrated memory- and chunking-based metric of comprehension complexity, in which we postulate that STM limitations of 7±2 (Miller, 1956) are more sensitive to the Instant Chunk Number (ICN), while WM limitations of 4±1 (Cowan, 2001) are more sensitive to the mean dependency distance (MDD; Liu, 2008) during language processing. Examples of concrete calculations of our new metric are presented vis-à-vis Liu’s MDD metric and related methodology issues are also discussed. We conclude the paper by echoing some recently repeated calls (O'Grady, 2012 & 2017; Gómez-Rodríguez et al., 2019; Wen, 2019) to include STM and WM limitations as part and parcel of the language device (LD; cf. Chomsky, 1957) in that their impacts are ubiquitous and permeating in all essential linguistic domains ranging from phonology to grammar, discourse comprehension and production.

Working memory in language learning and bilingual development

“…Similarly, Gomez-Rodriguez, Christiansen and Ferrer-i-Cancho (2019) by analyzing the big data of crosslinguistic corpora, argues that (working) memory limitations are hidden in grammar. Lu and Wen (2022) also set out to make the distinction between the magical number 4 (Cowan, 2001) and the magical number 7 (Miller, 1956) in that short-term memory limitations of 7 refer to the range of the momentary chunk number (MCN), while the working memory limitations of 4 refer to the average mean of momentary chunk number (MMCN; cf. Liu's MMD, 2008) within the 'focus of attention' analogous to Cowan's (1999) embedded processes model (see also Cowan, Morey, & Naveh-Benjamin, 2020).…”

“…Following an extensive review of decades of research, Cowan (2001) concluded that the fundamental capacity of the focus of attention is, on average, four chunks (i.e., groups) of items. Because individuals vary in the capacity of the focus of attention, the actual range is from three to five (i.e., the magical number of 4 plus or minus 1; also see Lu & Wen, 2022). The capacity-limited focus of attention is analogous to the central executive in the Baddeley and Hitch (1974;Baddeley, 1986) model of working memory, to the supervisory attentional system in Norman and Shallice's (1986) model, and controlled attention in Posner and Snyder (1975) and Schneider and Shiffrin's (1977) models.…”

Working Memory Models and Measures in Language and Bilingualism Research: Integrating Cognitive and Affective Perspectives

“…[ 88 ]), the role of WM or general memory as a whole is generally downplayed and marginalized, sometimes to the extent of negligibility (e.g., [ 89 , 90 ]; cf. [ 91 ]).…”

“…(p. 3) Contrary to Chomsky’s dominant view in mainstream linguistics, other processing- and performance-oriented theoretical linguists (e.g., [ 41 , 93 ]) as well as emergentist-oriented (e.g., [ 94 ]) and typological or dependency grammar–oriented linguists ([ 95 ]; cf. [ 96 ]), have all attached great emphasis to the role of WM, holding the view that WM limitations are part and parcel of the language parser (or the language device; [ 84 , 91 ]). It is even claimed that WM limitations play a pervasive, albeit sometimes ‘hidden’, role in key domains of language design, acquisition, and processing of linguistic structures and constructions ranging from phonology to grammar and discourse ([ 94 ]; cf.…”